Social Interaction Detection on SocialPulse
88.18PrecisionMulti-modal meta (NearestCentroid)
Evaluation Results
| Method | Links | ||||||
|---|---|---|---|---|---|---|---|
| Multi-modal meta (NearestCentroid)Modality=Multi-modal, Classifier=NearestCentroid, N=336982026.02 | 88.18 | 90.38 | 90.27 | 90.33 | 89.09 | 90.31 | |
| Audio only meta (Logit)Modality=Audio only, Classifier=Logit, N=336982026.02 | 88.17 | 90.54 | 90.2 | 90.37 | 89.1 | 90.31 | |
| Audio only meta (NearestCentroid)Modality=Audio only, Classifier=NearestCentroid, N=336982026.02 | 88.15 | 90.61 | 90.15 | 90.38 | 89.09 | 90.29 | |
| Multi-modal meta (Logit)Modality=Multi-modal, Classifier=Logit, N=336982026.02 | 88.02 | 90.59 | 90 | 90.3 | 88.97 | 90.19 | |
| Audio only meta (LinearSVC)Modality=Audio only, Classifier=LinearSVC, N=336982026.02 | 87.97 | 91.01 | 89.77 | 90.39 | 88.97 | 90.16 | |
| Multi-modal meta (LinearSVC)Modality=Multi-modal, Classifier=LinearSVC, N=336982026.02 | 87.86 | 91.17 | 89.56 | 90.36 | 88.88 | 90.07 | |
| Baseline 2 (YAMNet + FC)Backbone=YAMNet, Architecture=YAMNet + Fully Connected Layers, N=336982026.02 | 87.77 | 82.36 | 92.57 | 87.47 | 87.62 | 89.36 | |
| Baseline 1 (MobileNetV1)Backbone=MobileNetV1, N=336982026.02 | 82.93 | 71.18 | 91.07 | 81.13 | 81.92 | 84.81 |